Hybrid MapReduce Workflow
Yang Ruan, Zhenhua Guo, Yuduo Zhou, Judy Qiu, Geoffrey Fox
Indiana University, US
Outline
• Introduction and Background
– MapReduce
– Iterative MapReduce
– Distributed Workflow Management Systems
• Hybrid MapReduce (HyMR)
– Architecture
– Implementation
– Use case
• Experiments
– Performance
– Scaleup
– Fault tolerance
• Conclusions
MapReduce
• Introduced by Google
• Hadoop is an open source MapReduce framework
Mapper: read input data, emit key/value pairs
Map
User
Program
fork
assign
map
fork
fork
Master
Reducer: accept a key and all the values belongs to that key, emits final output
assign
reduce
Reduce
Input Data
Worker
Split 0
Split 1 read
Split 2
Worker
local
write
Worker
Worker
Worker
remote read, sort
write
Output
File 0
Output
File 1
Iterative MapReduce(Twister)
•
•
•
•
•
•
•
•
Iterative applications: K‐means, EM
An extension to MapReduce
Long‐running mappers and reducers.
Use data streaming instead of
file I/O
Keep static data in memory Use broadcast to send out updated data to all mappers
Use a pub/sub messaging infrastructure
Naturally support parallel iterative applications efficiently
Workflow Systems
• Traditional Workflow Systems
– Focused on dynamic resource allocation
– Pegasus, Kepler, Taverna
• MapReduce Workflow Systems
– Oozie
• Apache Project
• Use XML to describe workflows
– MRGIS
• Focus on GIS applications
– CloudWF
• Optimized for usage in Cloud
– All based on Hadoop
Why Hybrid?
• MapReduce
–
–
–
–
Lack of the support of parallel iterative applications
High overhead on iterative application execution
Strong fault tolerance support
File system support
• Iterative MapReduce
– No file system support, the data are saved in local disk or NFS
– Weak fault tolerance support
– Efficient iterative application execution
HyMR Architecture
• Concrete model
– Use PBS/TORQUE for resource allocation
– Focused on efficient workflow execution after resource is allocated
• User Interface
– WF definition in Script/XML
• Instance Controller
–
–
–
–
WF model: DAG
Manage workflow execution
Job status checker
Status updates in XML
Job and Runtime Controller
• Job Controller
–
–
–
–
Manage job execution
Single Node Job: File Distributor, File Partitioner
Multi‐Node Job: MapReduce Job, Iterative MapReduce Job
Twister Fault Checker: Detect faults and notify Instance Controller
• Runtime Controller
– Runtime Configuration: save the user from complicate Hadoop and Twister configuration and start the runtime automatically
– Persistent Runtime: reduce time cost of restarting runtimes once a job is finished
– Support Hadoop and Twister
File System Support in Twister
• Add HDFS support for Twister
– Before: explicit data staging phase
– After: implicit data staging as same as Hadoop
A Bioinfo Data Visualization Pipeline
• Input: FASTA File
• Output: A coordinates file contains the mapping result from dimension reduction
• 3 main components:
– Pairwise Sequence alignment: reads FASTA file, generates dissimilarity matrix
– Multidimensional Scaling(MDS): reads dissimilarity matrix, generates coordinates file
– Interpolation: reads FASTA file and coordinates file, generates final result
…
>SRR042317.123
CTGGCACGT…
>SRR042317.129
CTGGCACGT…
>SRR042317.145
CTGGCACGG…
…
Twister‐Pipeline
• Hadoop does not directly support MDS (iterative application). Incur high overhead
• All of the data staging are explicitly considered as a job
Hybrid‐Pipeline
• In HyMR pipeline, distributed data are stored in HDFS. No explicit data staging is needed as partitioned data are write into and read from HDFS directly.
Pairwise Sequence Alignment
Block (0,0)
Input Sample Fasta Partition 1
Block (0,1)
Input Sample FastaPartition 2
…
Block (n,0)
…
Input Sample Fasta Partition n
…
Block
(n‐1,n‐1)
Map
M
M
…
Reduce
Dissimilarity Matrix Partition 1
R
…
R
Block (0,1)
Block (0,2)
Block (0,n‐1)
Block (1,0)
Block (1,1)
Block (1,2)
Block (1,n‐1)
Block (2,0)
Block (2,1)
Block (2,2)
Block (2,n‐1)
Block Block (n‐1, 0) (n‐1, 1)
Block
(n‐1,n‐1)
…
Dissimilarity Matrix Partition n
M
Sample Data File I/O
Block (0,0)
Dissimilarity Matrix Partition 2
Network Communication
• Used for generating all‐pair dissimilarity matrix
• Use Smith‐Waterman as alignment algorithm
• Improve task granularity to reduce scheduling overhead Multidimensional Scaling (MDS)
• Scaling by Majorizing a Complicated Function (SMACOF)
• Two MapReduce Job in one iteration
Sample Data File I/O
Map
Sample Label File I/O
Map
Reduce
Network Communication
Reduce
Input Dissimilarity Matrix Partition 1
M
Input Dissimilarity Matrix Partition 2
M
…
…
…
Input Dissimilarity Matrix Partition n
M
M
Parallelized SMACOF Algorithm
Stress
Calculation
M
R
C
M
R
C
Sample Coordinates
MDS Interpolation
• SMACOF use O(N2) memory, which limits its applicability on large collection of data
• Interpolate out‐sample sequences into target dimension space by giving k nearest neighbor sample sequences’ mapping result
Input Sample Coordinates
Map
Reduce
Input Sample Fasta
Input Out‐Sample Fasta Partition 1
Input Out‐Sample Fasta Partition 2
…
Input Out‐Sample Fasta Partition n
M
M
R
…
…
R
C
Final Output
M
Sample Data File I/O
Out‐Sample Data File I/O
Network Communication
Experiment Setup
• PolarGrid cluster in Indiana University (8 cores per machine)
• 16S rRNA data from the NCBI database.
• Num. of sequences: from 6144 to 36864
• Sample set and out‐sample set: 50 – 50
• Node/Core number from 32/256 to 128/1024
Performance Comparison
Tested on 96 nodes, 768 cores
Differences increases when data size is larger
Write/read files to/from HDFS directly
Runtime starts take longer time
Execution includes read/write I/O, which is higher than local 8
disk
TimeCost(thousand
second)
•
•
•
•
•
7
6
5
4
3
2
1
0
Twister‐pipeline
Hybrid‐pipeline
6144 12288 18432 24576 30720 36864
Datasize
Detail Time Analysis
• Twister‐pipeline
– Data Staging time is longer when data size increases
– Less runtime start/stop time
100%
Twister‐pipeline
• Hybrid‐pipeline
– Data Staging time is fixed due to map task number is fixed
– Longer execution time
100%
Hybrid‐pipeline
80%
80%
60%
60%
40%
40%
20%
20%
0%
0%
6144 12288 18432 24576 30720 36864
Datasize
RuntimeControl Execution DataStaging
6144 12288 18432 24576 30720 36864
Datasize
RuntimeControl Execution DataStaging
Scale‐up Test
• Hybrid‐pipeline performs better when # of node increases
– Data distribution overhead from Twister increases
– Scheduling overhead for Hadoop increases, but not much
7
16
6
14
5
4
3
2
Twister‐speedup
Hybrid‐speedup
1
0
0
256
512
768
Corenumber
1024
TimeCost(thousand
seconds)
Speedup(hundreds)
• For pure computation time: Twister‐pipeline performs slightly better since all the files are in local disk when jobs are run
Twister‐execution‐time
Hybrid‐execution‐time
12
10
8
6
4
2
0
0
256
512
768
Corenumber
1024
Fault Tolerance Test
•
•
Fault tolerance, kill 1/10 nodes manually at different time during execution
10% and 25% are at PSA; 40% is at MDS; 55%, 70% and 85% are at Interpolation
If the node is killed when using Hadoop runtime, the tasks will be rescheduled immediately; Otherwise HyMR will restart the job
TimeCost(thousandseconds)
•
4
3.5
3
2.5
2
1.5
1
0.5
0
10%
Hybrid‐10nodes
Hybrid‐1node
Twister‐1node
25%
40%
55%
70%
Timepercentage
85%
Conclusions
• First hybrid workflow system based on MapReduce and iterative MapReduce runtimes
• Support iterative parallel application efficiently
• Fault tolerance and HDFS support added for Twister
Questions?
Supplement
Other iterative MapReduce runtimes
Haloop
Spark
Extension based on Hadoop Iterative MapReduce by keeping long running mappers and reducers
Task Scheduler keeps data locality for mappers and reducers
Input and output are cached on local disks to reduce I/O cost between
iterations
Build on Nexus, a cluster manger keep long running executor on each node. Static data are cached in memory between iterations.
Fault tolerance same as Use Resilient Distributed Hadoop. Dataset to ensure the fault Reconstruct cache to the tolerance
worker assigned with failed worker’s partition.
Pregel
Large scale iterative graphic processing framework
Use long living workers to keep the updated vertices between Super Steps.
Vertices update their status during each Super Step.
Use aggregator for global coordinates.
Keep check point through each Super Step. If one worker fail, all the other work will need to reverse.
Different Runtimes Comparison
Name
Iterative
Fault Tolerance
File System
Scheduling
Higher Caching
level language
Worker
Unit
Environment
Google
No
Strong
GFS
Dynamic
Sawzall
‐‐
Process
C++
Hadoop
No
Strong
HDFS
Dynamic
Pig
‐‐
Process
Java
Twister
Yes
Weak
‐‐
Static
‐‐
Memory
Thread
Java
Haloop
Yes
Strong
HDFS
Dynamic
‐‐
Disk
Process
Java
Spark
Yes
Weak
HDFS
Static
Scala
Memory
Thread
Java
Pregel
Yes
Weak
GFS
Static
‐‐
Memory
Process
C++